https://github.com/rhecosystemappeng/rhdh-templates
Repository to manage RHDH templates
https://github.com/rhecosystemappeng/rhdh-templates
Last synced: 5 months ago
JSON representation
Repository to manage RHDH templates
- Host: GitHub
- URL: https://github.com/rhecosystemappeng/rhdh-templates
- Owner: RHEcosystemAppEng
- License: apache-2.0
- Created: 2025-04-15T18:58:23.000Z (about 1 year ago)
- Default Branch: main
- Last Pushed: 2025-10-02T00:03:43.000Z (9 months ago)
- Last Synced: 2025-10-02T00:11:58.809Z (9 months ago)
- Size: 5.01 MB
- Stars: 2
- Watchers: 2
- Forks: 6
- Open Issues: 1
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
This guide provides step-by-step instructions for installing the Red Hat Golden Template path using the RHEcosystemAppEng/RHDH-templates repository.
---
## ✅ Prerequisites
Before getting started, ensure you have the following:
- **OpenShift CLI (oc)**: [Download and install](https://developers.redhat.com/learning/learn:openshift:download-and-install-red-hat-openshift-cli/resource/resources:download-and-install-oc) Openshift command-line interface
- **Platform Access**: Access to either [TAP](https://docs.redhat.com/en/documentation/red_hat_trusted_application_pipeline/1.0/html-single/installing_red_hat_trusted_application_pipeline/index) or a running RHDH instance. Helm Chart installation available [here](https://github.com/redhat-ai-dev/ai-rhdh-installer)
- **Hugging Face API Token**: A valid authentication token from [Hugging Face](https://huggingface.co/docs/hub/en/security-tokens)
---
### 🚀 Step-by-Step Instructions
### 1. Create a Kubernetes secret for HF token
## 🔓 Without Vault
Set up your Hugging Face authentication:
1. Configure your token as an environment variable:
```bash
export HF_TOKEN=
```
Replace with your actual Hugging Face API token.
2. Create the secret in your OpenShift namespace:
```bash
oc create secret generic huggingface-secret \
-n \
--from-literal=HF_TOKEN=$HF_TOKEN
```
Replace with the namespace where your RAG application is deployed.
---
### 🔐 With Vault + External Secrets Operator
> **Note**: Use this approach if you have Vault and External Secrets Operator configured in your cluster for centralized secret management.
1. **Access Vault UI**:
```bash
# Get the Vault route
oc get route -n vault
# Get the Vault token
oc get secret -n vault vault-token -o jsonpath="{.data.token}" | base64 --decode
```
Open the Vault route in your browser and log in using the token method with the retrieved token.
2. **Create the secret in Vault**:
- Select the **KV** secret engine
- Navigate to: `secret/`
- Set path as: `secrets/ai-kickstart`
- Click **Create secret** (Shown on image 1)
- Add secret data:
- **Key**: `hf_token`
- **Value**: ``
- Click **Save** (Shown on image 2)
(1)
(2)
> **Note**: **The ExternalSecret Operator will map `hf_token` → Kubernetes key `HF_TOKEN`**
---
### 2. Login to Developer Hub
* Sign in to Developer Hub via GitLab using your GitLab credentials
---
### 3. Register AI Templates
1. **Navigate to Create**:
- From the Developer Hub sidebar, click **"Create"**
2. **Register templates**:
- Click **"Register Existing Component"**
3. **Import the template repository**:
- Paste this URL into the input field:
```
https://github.com/RHEcosystemAppEng/RHDH-templates/blob/main/showcase-templates.yaml
```
- Click **"Analyze"**
- Click **"Import"** to complete registration
---
### 4. Available Templates
Once registered, you'll see these AI-powered templates in the Catalog->Template page:
- **🤖 RAG Chatbot Kickstart** (`chatbot-rag-kickstart-template`)
Deploy a complete RAG (Retrieval Augmented Generation) architecture using LLaMA Stack, OpenShift AI, and PGVector. Includes document ingestion pipeline and vector database for intelligent question-answering.
- **🎯 AI Virtual Agent** (`ai-virtual-agent-kickstart-template`)
Create an intelligent virtual assistant powered by OpenShift AI and PGVector. Perfect for building conversational AI applications with advanced reasoning capabilities.
- **📊 AI Metrics Summarizer** (`ai-metric-summarizer-kickstart-template`)
Build a specialized chatbot that analyzes AI model performance metrics from Prometheus and generates human-readable summaries using LLaMA models. Ideal for AI observability and monitoring.
---
### 5. Launch a Template
Once you've registered the templates, follow these steps to deploy an AI application:
#### **Navigate to Self-Service Catalog**
- From the Developer Hub sidebar, click **"Create"**
- You'll see the available AI templates listed
#### **Choose Your Template**
Select one of the registered templates:
- **Chatbot-Rag Kickstart** - for RAG document-based Q&A systems
- **AI Virtual Agent** - for conversational AI assistants
- **AI Metrics Summarizer** - for AI observability and monitoring
#### **Configure Template Parameters**
Fill in the guided form with your specifications:
**Application Information:**
- **Name**: Unique identifier for your component (e.g., `my-ai-chatbot`)
- **Description**: Brief description of your application
**Repository Details:**
- **Host Type**: Choose GitHub or GitLab
- **Repository Owner**: Your organization name
- **Repository Name**: Name for the source repository
- **Namespace**: Kubernetes namespace for deployment
**AI Model Configuration:**
- **Language Model**: Select from available LLaMA variants
- **Safety Model**: Optional LLaMA Guard for content filtering
- **GPU Tolerance**: Configure hardware requirements
#### **Review and Create**
- Review all configured parameters
- Click **"Review"** to validate your inputs
- Click **"Create"** to initiate the template deployment
#### **Automatic Deployment Process**
The template will automatically:
1. **Build** the software component with your specifications
2. **Publish** source and GitOps repositories to your chosen platform
3. **Register** the component in the Developer Hub catalog
4. **Deploy** via ArgoCD using GitOps workflows
#### **Access Your Application**
Once complete, use the provided links to:
- View source repository
- Monitor GitOps deployment
- Access the component in the catalog
- Review ArgoCD applications
#### **Post-Deployment Steps:**
Once your application is deployed, you'll need to make the following changes in your GitOps repository:
1. Uncomment the Toolhive configuration to enable the service
2. Delete the chart.lock file to allow Helm to regenerate dependencies
3. Commit these changes to trigger the GitOps sync